Michael J. Rothman

Development and validation of a continuous measure of patient condition using the Electronic Medical Record

Patient condition is a key element in communication between clinicians. However, there is no generally accepted definition of patient condition that is independent of diagnosis and that spans acuity levels. We report the development and validation of a continuous measure of general patient condition that is independent of diagnosis, and that can be used for medical-surgical as well as critical care patients. A survey of Electronic Medical Record data identified common, frequently collected non-static candidate variables as the basis for a general, continuously updated patient condition score. We used a new methodology to estimate in-hospital risk associated with each of these variables. A risk function for each candidate input was computed by comparing the final pre-discharge measurements with 1-year post-discharge mortality. Step-wise logistic regression of the variables against 1-year mortality was used to determine the importance of each variable. The final set of selected variables consisted of 26 clinical measurements from four categories: nursing assessments, vital signs, laboratory results and cardiac rhythms. We then constructed a heuristic model quantifying patient condition (overall risk) by summing the single-variable risks. The models validity was assessed against outcomes from 170,000 medical-surgical and critical care patients, using data from three US hospitals. Outcome validation across hospitals yields an area under the receiver operating characteristic curve(AUC) of ≥0.92 when separating hospice/deceased from all other discharge categories, an AUC of ≥0.93 when predicting 24-h mortality and an AUC of 0.62 when predicting 30-day readmissions. Correspondence with outcomes reflective of patient condition across the acuity spectrum indicates utility in both medical-surgical units and critical care units. The model output, which we call the Rothman Index, may provide clinicians with a longitudinal view of patient condition to help address known challenges in caregiver communication, continuity of care, and earlier detection of acuity trends.

Placing clinical variables on a common linear scale of empirically based risk as a step towards construction of a general patient acuity score from the electronic health record: a modelling study

Objective To explore the hypothesis that placing clinical variables of differing metrics on a common linear scale of all-cause postdischarge mortality provides risk functions that are directly correlated with in-hospital mortality risk. Design Modelling study. Setting An 805-bed community hospital in the southeastern USA. Participants 42302 inpatients admitted for any reason, excluding obstetrics, paediatric and psychiatric patients. Outcome measures All-cause in-hospital and postdischarge mortalities, and associated correlations. Results Pearson correlation coefficients comparing in-hospital risks with postdischarge risks for creatinine, heart rate and a set of 12 nursing assessments are 0.920, 0.922 and 0.892, respectively. Correlation between postdischarge risk heart rate and the Modified Early Warning System (MEWS) component for heart rate is 0.855. The minimal excess risk values for creatinine and heart rate roughly correspond to the normal reference ranges. We also provide the risks for values outside that range, independent of expert opinion or a regression model. By summing risk functions, a first-approximation patient risk score is created, which correctly ranks 6 discharge categories by average mortality with p<0.001 for differences in category means, and Tukeys Honestly Significant Difference Test confirmed that the means were all different at the 95% confidence level. Conclusions Quantitative or categorical clinical variables can be transformed into risk functions that correlate well with in-hospital risk. This methodology provides an empirical way to assess inpatient risk from data available in the Electronic Health Record. With just the variables in this paper, we achieve a risk score that correlates with discharge disposition. This is the first step towards creation of a universal measure of patient condition that reflects a generally applicable set of health-related risks. More importantly, we believe that our approach opens the door to a way of exploring and resolving many issues in patient assessment.

Sepsis as 2 problems: Identifying sepsis at admission and predicting onset in the hospital using an electronic medical record–based acuity score ☆ ☆☆

Purpose: Early identification and treatment improve outcomes for patients with sepsis. Current screening tools are limited. We present a new approach, recognizing that sepsis patients comprise 2 distinct and unequal populations: patients with sepsis present on admission (85%) and patients who develop sepsis in the hospital (15%) with mortality rates of 12% and 35%, respectively. Methods: Models are developed and tested based on 258 836 adult inpatient records from 4 hospitals. A “present on admission” model identifies patients admitted to a hospital with sepsis, and a “not present on admission” model predicts postadmission onset. Inputs include common clinical measurements and the Rothman Index. Sepsis was determined using International Classification of Diseases, Ninth Revision, codes. Results: For sepsis present on admission, area under the curves ranged from 0.87 to 0.91. Operating points chosen to yield 75% and 50% sensitivity achieve positive predictive values of 17% to 25% and 29% to 40%, respectively. For sepsis not present on admission, at 65% sensitivity, positive predictive values ranged from 10% to 20% across hospitals. Conclusions: This approach yields good to excellent discriminatory performance among adult inpatients for predicting sepsis present on admission or developed within the hospital and may aid in the timely delivery of care. HighlightsThere are 2 unequal but distinct sepsis populations.Eighty‐five percent of sepsis patients arrive with sepsis; mortality: 12%.Fifteen percent of sepsis patients develop sepsis in the hospital; mortality: 35%.Two models built using the RI identify and predict onset.Models flag 65% to 75% of sepsis patients with reasonable PPVs.